Details
-
Bug
-
Status: Open (View Workflow)
-
Major
-
Resolution: Unresolved
-
10.5, 10.6, 10.11
-
None
-
None
Description
The easiest way to trigger this is by disabling the GBH in 10.11+:
# ... (setup)
|
set spider_same_server_link= 1; |
set global spider_same_server_link= 1; |
set spider_select_column_mode= 0; |
set spider_disable_group_by_handler= 1; |
evalp CREATE SERVER srv FOREIGN DATA WRAPPER mysql |
OPTIONS (SOCKET "$MASTER_1_MYSOCK", DATABASE 'test',user 'root'); |
CREATE TABLE t1 (c int); |
CREATE TABLE t1_s (c int) |
ENGINE=SPIDER COMMENT='wrapper "mariadb", srv "srv", table "t1"' ; |
SELECT MAX(c) FROM t1_s; |
drop table t1_s, t1; |
drop server srv; |
# ... (cleanup)
|
Running this test results in
mysqltest: At line 16: query 'SELECT MAX(c) FROM t1_s' failed: ER_BAD_FIELD_ERROR (1054): Unknown column 'c' in 'SELECT'
|
The log shows that the SELECT query is translated to select max(`c`),min(`c`), missing the FROM part.
The problem lies in ha_spider::index_first_internal, which is untested in the current spider regression test suites. That is, the following change causes no regression:
modified storage/spider/ha_spider.cc
|
@@ -1911,6 +1911,7 @@ int ha_spider::index_prev(
|
int ha_spider::index_first_internal(
|
uchar *buf
|
) {
|
+ abort();
|
int error_num;
|
backup_error_status();
|
DBUG_ENTER("ha_spider::index_first_internal"); |
The bug is the same as the failure of spider/bugfix.mdev_35874 with --view-protocol and the current draft patch of MDEV-36307 applied (bb-10.5-mdev-35452 d9d2fb9d07e47296faee79d230a38f7f9a0b3ef6)
Attachments
Issue Links
- is blocked by
-
-
- Open
-
- is part of
-
-
- Open
-
- relates to
-
-
- Open
-